EP2273229A1 - Method for determining the 3D coordinates of an object - Google Patents

Abstract

The method involves surrounding an object (1) by a set of reference probes (2-6) that include coded reference marks (7, 8). A set of shots (10-14) are produced by the object in such a manner that a part of the object and a part of one probe are enclosed. The probes are formed longer than a measurement field of a three dimensional sensor (9), where the probes are designed in bar shape. The reference marks are automatically decoded. The shots are linked to one another by a three dimensional matching process. The reference marks are provided on rear sides of the probes.

Description

Method for determining the 3D coordinates of an object

The invention relates to a method for determining the 3D coordinates of an object.

The EP 553 266 B1 discloses a method for determining the 3D coordinates of an object that includes a scanner through which the object is scannable. The scanner determines the 3D coordinates in the scanner reference system. There is also a tracking system that can determine the position and direction of the scanner. From a computer, the 3D coordinates of the object in the absolute frame of reference can be determined and recorded from the 3D coordinates in the scanner reference system and from the data of the position and direction of the scanner in the absolute reference frame. The scanner may be held such that it is independently movable with respect to the tracking system and the object.

From the US 5,805,289 For example, a method for determining the 3D coordinates of an object is known in which reference scenes are attached to an object. With a photogrammetry method absolute 3D coordinates are determined.

The DE 198 40 334 A1 discloses a device for use as a navigational backdrop in the measurement of objects with optical three-dimensional coordinate measuring technique. The navigation gate contains a component with a grid structure, on which several measuring marks are mounted. On the side facing away from the component of the navigation backdrop a fastening device for attachment to the object is mounted.

The WO 01/88471 A1 discloses a method for determining the 3D shape of an object, wherein the object is connected to a reference body having reference objects. The object and the reference body are scanned by an opto-electronic sensor, from the measurement signals of the 3D coordinates of the surface points are calculated.

From the EP 1 724 549 A2 For example, a method for determining the 3D coordinates of an object is known in which the 3D coordinates of a partial surface of the object are determined by a 3D measuring device whose position is determined by a tracking system. Subsequently, the 3D coordinates of a partially overlapping, adjacent partial surface of the object are determined in this way. The 3D coordinates of the overlap area are assembled by a matching method.

The EP 2 034 269 A1 discloses a method for three-dimensional digitization of objects with a 3D sensor comprising a projector and one or more cameras. The projector projects a pattern onto the object captured by the cameras.

The object of the invention is to propose an improved method for determining the 3D coordinates of an object.

According to the invention this object is achieved by the characterizing features of claim 1. From the object, which is surrounded by several Referenzkulissen with coded reference marks, several recordings are made. These recordings are made in such a way that each part of the object and a part of a reference scenery is contained on it.

The method according to the invention can be carried out on the basis of all known methods for determining the 3D coordinates of an object. Particularly suitable are optical methods. To carry out these optical methods An optical 3D sensor can be used. The optical 3D sensor may include a camera, which may include an optical system and / or a planar sensor, in particular a CCD sensor. Furthermore, the 3D sensor may include one or more light sources.

It is advantageous if the position of the 3D sensor can be determined, in particular by a tracking system. In this way, it is possible that the 3D sensor can determine the relative 3D coordinates of the object in the frame of reference of the 3D sensor, and that from these and the absolute coordinates of the 3D sensor, the absolute 3D coordinates of the object can be determined. Examples are in the EP 0 553 266 B1 , of the EP 1 724 549 A2 and the EP 2 034 269 A1 to which reference is made.

The 3D optical sensor may comprise a fringe projection system, in particular a white light fringe projection system. From such a 3D sensor, a stripe pattern is projected onto the object. Furthermore, the object is recorded. The recordings are evaluated. The recordings and the evaluations can be displayed and / or saved.

When determining the 3D coordinates, which can also be referred to as 3D digitizing or measuring the object, with a white-light fringe projection system and also with other systems, it is necessary to move the 3D sensor in space in different recording positions, if the object to be measured is larger than the measuring field of the 3D sensor. The individual shots of the object must then be put together. For this it is necessary to register the individual recordings in a higher-level coordinate system. The superordinate coordinate system is preferably an absolute coordinate system, for example the coordinate system of the room in which the object is located and in which the measurement of the object takes place. This so-called global registration requires high accuracy.

• die Bauart der Vollkulissen orientiert sich hinsichtlich ihrer Größe und Gestalt an den zu vermessenden Objekten, so daß die Vollkulissen bei großen Meßobjekten teuer und unhandlich sind;
• für jedes Meßobjekt muß eine eigene Kulisse hergestellt und fotogrammetrisch eingemessen werden, so daß diese Kulissen bei einer Vielzahl unterschiedlicher Meßobjekte eine beträchtlichen Kostenfaktor darstellen;
• die Ansprüche an die Stabilität der Vollkulisse sind sehr hoch, sie muß also im Ganzen mechanisch sehr stabil ausgelegt sein, was Gewicht und Kosten erhöht;
• wenn der zu messende Objekttyp wechselt, muß der komplette Meßaufbau einschließlich der unhandlichen Vollkulisse umgebaut werden;
• das regelmäßige fotogrammetrische Einmessen einer Vollkulisse zur Sicherstellung und Überwachung von deren Genauigkeit ist größenbedingt und gewichtsbedingt sehr aufwendig.

In the case of white-light fringe projection, the composition of the individual images in a higher-level coordinate system is typically carried out by means of photogrammetry, reference marks having to be applied to the object or on a reference surface connected to the object. The stamps must be measured separately with photogrammetry equipment. The method of photogrammetry offers high accuracy, but has the disadvantage that for the preparation of the object with reference marks and for their separate measurement, an additional effort is required. Furthermore, it is necessary to produce the reference object adapted to the object to be measured on which reference marks are applied, which must be measured photogrammetrically. The scenes adapted to the object to be measured must be placed as close to and around the object as possible. Although these scenes, which can also be referred to as full scenes, allow the measurement of the object with high accuracy, they have the following disadvantages:

• the design of the full scenes is oriented in terms of their size and shape of the objects to be measured, so that the full scenes are expensive and unwieldy in large objects to be measured;
• for each DUT must have its own backdrop produced and measured photogrammetrically, so that these scenes represent a considerable cost factor in a variety of different DUTs;
• the demands on the stability of the full scenery are very high, so it must be designed as a whole mechanically very stable, which increases weight and cost;
• if the object type to be measured changes, the complete measurement setup including the unwieldy full scenery must be rebuilt;
• the regular photogrammetric calibration of a full set to ensure and monitor their accuracy is very large due to the size and weight.

These disadvantages are avoided by the use of multiple reference scenes, which can also be referred to as Referenzkulissenmodule. Through the use of multiple reference scenes a much simpler, cheaper and more variable measurement setup is possible, which nevertheless ensures a high accuracy of the measurement results.

The reference scenes do not have to be attached to the object. It is possible to arrange the reference scenes at a distance from the object.

The reference marks can be coded in themselves. However, it is also possible to use reference marks which are not coded in themselves, but which contain a coding due to their relative arrangement to each other. For self-coded reference marks, it is sufficient to record this reference mark. Of measurement marks that are not encoded in itself, such a high number must be included in a recording that from the relative position of these measurement marks each other, the coding can be decoded.

The shots of the object may overlap. The object can be measured section by section.

The part of the Referenzkulisse contained in the recording, preferably comprises at least one coded reference mark. Preferably, such a part of a reference scenery is recorded, from which the position and / or orientation of the Referenzkulisse can be determined.

It is also advantageous if the reference scenes are calibrated. This is preferably done by the position of all reference marks on the reference gate is determined in advance. Preferably, the data of the position of all reference marks, based on the respective Referenzkulisse stored.

Advantageous developments are described in the subclaims.

It is advantageous if some or all Referenzkulissen are longer than the measuring field of the 3D sensor. The reference scenes are thus longer than the size of the recorded during a recording area of the object. This results in advantages for large objects, ie objects for the measurement of several shots are required.

Some or all reference scenes may be rod-shaped.

It is advantageous if recordings which contain a part, in particular an end of a reference gate, contain a part, in particular an end, of a further reference gate. This is particularly advantageous if the reference scenes are rod-shaped. The further reference scenery may be adjacent to the first reference scenery. By capturing two reference scenes, it is possible to link the reference scenes together. For this purpose, it is advantageous if in each case a part of both reference scenes is recorded, which contains at least one reference mark and / or a part from which the position and / or orientation of the respective reference gate and / or their reference marks can be determined.

The reference marks can be decoded automatically. However, it is also possible to manually decode the reference marks. By decoding the reference marks can be assigned to a specific reference scenery, which can be identified with it.

Some or all images can be linked together using a 3D matching process to further increase accuracy.

It is advantageous if encoded reference marks are also provided on the backs of some or all of the reference scenes. Then the back side of the object can also be measured with the same measuring setup.

A further advantageous development is characterized in that in a first Meßdurchlauf the positions of the reference marks are detected and stored. The first measurement run can be performed without measuring the object. However, it is also possible to measure an object in this first measurement run. The detected positions of the reference marks can be transformed into the higher-level coordinate system. The detected and / or transformed positions can be stored.

It is advantageous if further Meßdurchläufe be carried out on the basis of stored positions of the reference marks. This is particularly advantageous if the positions of the reference marks are detected and optionally transformed and stored in a first measurement run. By using the stored positions of the reference marks measuring time can be saved.

Fig. 1

einen Meßaufbau mit einem zu vermessenden Objekt und mehreren Referenzkulissen in einer schematischen Ansicht von vorne,

Fig. 2.

den Meßaufbau gemäß Fig. 1 in einer Ansicht von hinten,

Fig. 3

den Meßaufbau gemäß Fig. 1 in einer abgewandelten Darstel- lung und

Fig. 4

verschiedene Arten von Referenzmarken.

An embodiment of the invention will be explained below with reference to the accompanying drawings. In the drawing shows

Fig. 1

a measuring structure with an object to be measured and several reference scenes in a schematic view from the front,

Fig. 2.

the measuring structure according to Fig. 1 in a view from behind,

Fig. 3

the measuring structure according to Fig. 1 in a modified representation and

Fig. 4

different types of reference marks.

The in Fig. 1 shown measurement structure is used to determine the 3D coordinates of the front of an object 1, namely a motor vehicle door (bodyshell door). The object 1 is surrounded by five reference scenes 2, 3, 4, 5, 6. Each Referenzkulisse 2-6 has reference marks, namely reference marks 7, which are encoded in itself, and reference marks 8, which are not encoded in itself, but which are arranged spatially relative to each other, that this spatial arrangement includes a coding.

The reference scenes 2-6 are rod-shaped.

For the production of images of the object, a 3D sensor 9 is present, which comprises a white-light fringe projection system and a camera with an optical system and a CCD sensor. Furthermore, a device, namely a tracking system, is provided, with which the position and orientation of the 3D sensor 9 in a higher-level coordinate system can be determined (not shown in the drawing). In this way, the absolute coordinates of the object 1 and the reference scenes 2-6 can be determined.

The reference scenes 2 - 6 have been measured. The result of this survey is stored in a control system which may include a computer, in particular a personal computer (not shown in the drawing). The result of the measurement of the reference scenes 2 - 6 comprises an unambiguous assignment of the respective reference marks 7, 8 to a specific reference gate 2 - 6, which can be identified in this way, and the position coordinates of the respective reference marks 7, 8 on the respective reference gate 2 - 6th

The reference scenes 2 - 6 are not attached to the object 1. They are placed in the vicinity of the object 1, preferably in such a way that they surround the object 1, as in FIG Fig. 1 shown. For this purpose reference objects with different lengths can be provided, namely reference objects 2, 3, 6, which are longer than the length 1 of the measuring field of the 3D sensor, and short reference scenes 4, 5, which are shorter than this length I. The reference scenes 2, 3, 6 are thus longer than the size I of the recorded during a recording area of the object. 2

The reference scenes 2 - 6 are in a not exactly calibrated position. They surround the object 1, but first they do not have to be positioned at exactly predetermined locations.

From the object 1 several shots are made by the 3D sensor 9. This is in Fig. 3 shown where the measuring fields of successive recordings 10, 11, 12, 13, 14 are shown. Each of the receptacles 10-14 includes a portion of the object 1 and a portion of a reference gallery 2-6. Further, each of the receptacles 10-14 includes a portion including an end of a first reference gate and a portion including an end of another reference gate. The receptacle 10 contains a part, namely an end of the reference gate 2 and a part, namely an end, the reference gate 3. The receptacle 11 contains parts or ends of the reference scenes 3 and 4. The receptacle 12 contains parts or ends of the reference scenes. 4 and 5. The receptacle 13 contains parts or ends of the reference scenes 5 and 6. The receptacle 14 contains parts or ends of the reference scenes 6 and 2. It is alledings also possible that recordings contain only a part of only one reference scenery (in the Drawing not shown).

The recordings partially overlap. The recordings 10 and 11 overlap, as do the recordings 13 and 14. The record 12 does not overlap with further recordings. Overlapping recordings can be linked together. If shots do not overlap, the 3D coordinates of the areas between shots will not be determined. These areas may be non-interest areas.

Each record contains reference marks 7, 8. The assignment of the size of the recordings and the distribution of the reference marks 7, 8 is such that with each shot 10 - 14 of each reference gate 2 - 6 whose position and orientation can be determined. For example, the receptacle 10 contains a self-coded reference mark 7 and four non-self-coded reference marks 8, although these non-self-coded reference marks 8 can be decoded by the position of their positions. In the same way, the receptacle 10 contains a self-coded reference mark 7 and two non-self-coded reference marks 8 of the reference gate 2. The same applies to the other recordings 11 to 14 and the remaining reference scenes 4 to 6.

In a first Meßdurchlauf the positions of the reference marks 7, 8 are detected and stored. During this first measurement run, the 3D coordinates of the object 1 can also be determined.

If subsequently further objects are to be measured, the stored positions of the reference marks 7, 8 can be used for this purpose. It is thus possible to remove the object 1, wherein the position of the reference scenes 2 - 6 to each other and thus the position of the reference marks 7, 8 is not changed. Subsequently, another, similar DUT is placed in a similar manner within the reference scenes 2-6. This further object can then be measured. This further measurement run and still further Meßdurchläufe can be performed on the basis of the stored positions of the reference marks 7, 8. Characterized in that the stored positions of the reference marks 7, 8 are used, the measuring time can be shortened.

How out Fig. 2 As can be seen, coded reference marks are also provided on the rear sides of the reference scenes 2-6, namely self-coded reference marks 7 'and non-self-coded reference marks 8' which are coded relative to each other by their spatial position. In this way, it is possible to measure the back of the object 1 with the same Meßaufbau. In particular, the described first-time measurement run of the front sides of the reference scenes 2 - 6 for complete acquisition and storage of the reference marks 7 ', 8' on the backsides of the reference scenes 2 - 6 is sufficient. For this purpose, the reference marks 7 ', 8' on the backs of the reference scenes 2-6 have been measured in advance, both in terms of their positions relative to each other and with respect to their positions relative to the reference marks 7, 8 on the front of the respective reference gate 2-6.

Fig. 4 shows various embodiments of self-coded reference marks 7, 7 '. Each reference mark comprises an unchanged non-coding element and coding elements. The non-coding element may consist of a circle 15 or a square 16 or a rectangle. This element determines the position of the respective reference mark. The coding elements may consist of a circular bar code 17, a dot pattern 18 of round or square dots, segment sections 19 or a pattern of dots 20 and dashes 21. The coding parts allow identification of the respective reference mark.

According to the invention, several reference scenes are used, which can be flexibly combined and which are arranged around the object to be measured before the measurement. During the measurement of the object and possibly other objects, the reference scenes are then no longer changed in their position. With a set of reference scenes, a variety of different DUTs can be measured.

Coded reference marks are attached to the reference scenes. They may be photogrammetry tags encoded within themselves and photogrammetry tags that are not encoded in themselves but by their position relative to each other. The reference marks on the Referenzkulissen are measured, for example by means of photogrammetry. In this way, there is an exact and inherently stable 3D point list for every reference backdrop. The photogrammetric measurement of each reference frame is only necessary once. The once measured point list is saved.

The object can be measured in sections with the 3D sensor, wherein in each individual image of the 3D sensor both a part of the object and a part of the reference marks are detected on one or more reference scenes. The reference scenes are at least partially designed in their length so that they are significantly larger than the measuring field of the 3D sensor, so that multiple adjacent shots of the 3D sensor can be linked via a Referenzkulisse and thus achieves high accuracy even when linking multiple measuring fields becomes.

The individual reference scenes are identified automatically (automatically) or manually in the measuring scene of the 3D sensor either via the analysis of the geometric arrangement of the non-self-coded reference marks or with the aid of the self-coded reference marks, which enables easier and safer identification.

After the first measurement, it is checked which reference gate or which reference scenes were or were detected next to the object in the respective recording, namely by the evaluation of the self-coded reference marks and / or by the evaluation or analysis of the geometric arrangement of not encoded reference marks. Subsequently, the respective 3D point list is transformed into the coordinate system of the first measurement. Any further measurement can now be transformed into this coordinate system, provided that reference marks of an already transformed reference scenery are detected. After each measurement, it is checked whether further, previously not yet recorded reference marks and / or reference scenes have been measured. If necessary, these are then transformed into the measurement scene. To increase the accuracy of the complete association of all measured object sections, so the complete association of all recordings, together with the detected reference scenes and / or reference marks via a 3D matching method are compared to each other.

By applying reference marks on all sides of the reference scenes, and thus also on their back, it is possible to measure the front and the back of an object in a common coordinate system.

After completion of the measurement on the object and thus present combined and aligned points list of all Referenzkulissen and their reference marks, it is mögilch for repeat measurements on other objects same Tys invite the complete point list of reference scenes and / reference marks completely at the start of the measurement sequence completely To save the transformation process of the individual reference scenes and / or reference marks during the repeat measurements. This presupposes that the position of the reference scenes for the repeat measurement has not been changed.

Claims (9)

Method for determining the 3D coordinates of an object (1),
characterized,
that the object (1) from a plurality of reference links (2, 3, 4, 5, 6) with the coded reference marks (7, 8) is surrounded
and in that a plurality of receptacles (10, 11, 12, 13, 14) of the object (1) are produced in such a way that a part of the object (1) and a part of a reference slide (2 - 6) are respectively contained thereon. Method according to Claim 1, characterized in that some or all of the reference scenes (2, 3, 6) are longer than the measuring field of the 3D sensor (9). Method according to claim 1 or 2, characterized in that some or all reference scenes (2-6) are rod-shaped. Method according to one of the preceding claims, characterized in that recordings which contain a part, in particular an end of a reference slide (2 - 6), contain a part, in particular an end of a further reference slide (2-6). Method according to one of the preceding claims, characterized in that the reference marks (7, 8) are automatically decoded. Method according to one of the preceding claims, characterized in that some or all recordings (10-14) are linked together by a 3D matching method. Method according to one of the preceding claims, characterized in that coded reference marks (7 ', 8') are also provided on the rear sides of some or all of the reference scenes (2-6). Method according to one of the preceding claims, characterized in that the positions of the reference marks (7, 8) are detected and stored in a first measuring pass. Method according to one of the preceding claims, characterized in that further measurement runs are carried out on the basis of stored positions of the reference marks (7, 8).

Images